fre TQMP Tutorials in Quantitative Methods for Psychology 1913-4126 2011-06-08 7 2 19 20 article Nathalie Loye nathalie.loye@umontreal.ca 1 Université de Montréal This special issue bears on the second colloque en méthodes quantitatives et sciences humaines, held june 2010 at Université de Montréal. http://www.tqmp.org/Content/vol07-2/p019/p019.pdf editorial special issue eng TQMP Tutorials in Quantitative Methods for Psychology 1913-4126 2011-06-08 7 2 21 31 article Frédéric Dandurand frederic.dandurand@umontreal.ca 1 Thomas R. Shultz thomas.shultz@mcgill.ca 2 Université de Montréal McGill University Automatic Maxima Detection is a statistical tool to detect and measure local maxima in functional data (Dandurand and Shultz, 2010). A typical use of AMD is the identification of growth spurts in time-varying data. The current paper provides a complete hands-on tutorial for how to use AMD, a new graphical user interface (GUI) and advanced scripts. http://www.tqmp.org/Content/vol07-2/p021/p021.pdf maxima detection graphical interface eng TQMP Tutorials in Quantitative Methods for Psychology 1913-4126 2011-06-08 7 2 32 41 article Léon Harvey leon_harvey@uqar.ca 1 Université du Québec à Rimouski This paper introduces Hidden Markov Models for the analysis of authentic learning data from an applied field. For illustrative purposes, it shows how classical 2-state all-or-none models can be extended to adequately fit the competence development process of nursery apprentices in a clinical context. It also presents some of the main underlying ideas, such as model specifications, parameters estimation, model selection, the Viterbi algorithm, and goodness-of-fit issues. http://www.tqmp.org/Content/vol07-2/p032/p032.pdf learning situation hidden markov model fre TQMP Tutorials in Quantitative Methods for Psychology 1913-4126 2011-06-08 7 2 42 53 article Louis Laurencelle louis.laurencelle@uqrt.ca 1 Stéphane Germain stephane.germain@ulaval.ca 2 Université du Québec à Trois-Rivières Université Laval In the parametric setting of item response theory (IRT), several procedures have been put up to estimate the examinee’s ability level (theta), some based solely on the information in the response protocol while others are grounded on a restrictive population model. Few studies however have endeavoured to compare the relative merits of these procedures. By way of an extensive Monte Carlo experiment, we studied the bias, precision and capture efficiency of four ability estimators : ML, BME, EAP, WARM, to which we added a winsorized ML estimator (WINS) and an extrapoled one (EXT), both enabling ML estimation for perfect (all items passed) and null (all items failed) scores. Dans le modèle paramétrique de la théorie des réponses aux items (TRI), l’estimation du niveau d’aptitude (theta) du répondant a fait l’objet de plusieurs procédures différentes, certaines basées uniquement sur le protocole de réponses observé et d’autres, appuyées sur un modèle de population contraignant. Peu d’études cependant ont entrepris d’établir les mérites comparatifs de ces procédures. Au moyen d’une large expérimentation Monte Carlo, nous étudions le biais, la précision et l’efficacité de capture de quatre estimateurs de capacité : MV, BME, EAP, WARM, auxquels nous ajoutons un estimateur MV winsorisé (WINS) et un autre extrapolé (EXT), qui permettent tous deux l’utilisation de l’estimateur MV pour les scores parfaits (tous items réussis) et nuls (tous items échoués). http://www.tqmp.org/Content/vol07-2/p042/p042.pdf item reponse theory monte carlo